Project Description

This thesis presents a new optimization-based approach to the long-term planning of interurban road networks. The traditional approach to the road network planning (or design) problem focuses on defining the investment decisions that will optimize the network efficiency under a given budget. The investment decisions can consist of either improving existing roads or adding new roads to an existing road network.

In the proposed approach, a multi-objective perspective is given to this planning problem by adding equity and robustness objectives to the efficiency objectives dealt with in the vast majority of the literature. The equity objective was included to reflect sustainable development concerns, aiming at a more balanced distribution of road investment benefits across the territory. The robustness objective was included in the approach to provide the network with the capability to respond to travel demand and infrastructure supply variations that will occur throughout its lifetime. Three to four measures were tested and compared for each of the three optimization objectives.

The proposed planning approach is consistent with the planning framework of the Highway Capacity Manual, using the concept of level of service (LOS) to assess traffic flow conditions. Decisions are taken according to a set of road types (or levels). That is, road investment decisions are defined in accordance with a hierarchy of road types. Each type of road is previously associated with a minimum LOS, which must express the minimum traffic conditions required by road authorities for that type of road. The optimal solution obtained with the proposed approach guarantees the minimum LOS for all roads in the network.

Another important feature of the approach is the assumption that travel demand is elastic, which means that in the long-term both traffic generation and trip distribution are influenced by road investment decisions. An unconstrained gravity model was used to calculate the expected number of trips between two traffic generation centers. It is assumed that the number of trips between a pair of centers is proportional to the size of the centers (population, employment, etc.) and inversely proportional to the (generalized) travel cost between the centers. Drivers were assumed to choose the least-cost path for their trips by considering the speeds corresponding to the minimum LOS to be verified for the roads included in the path.

The proposed approach relies on a mixed-integer non-linear bi-level optimization model that is difficult to solve for the global optimum. Therefore, three heuristic algorithms were developed to tackle it - a local search algorithm, a variable neighborhood search algorithm, and an enhanced genetic algorithm. The efficiency of the heuristics was tested from the standpoint of solution quality and computation effort.

Finally, the thesis presents OptRoad, a user-friendly computer program that was developed throughout this work and that was used to make all the calculations involved in this thesis.

Research Team

CITTA

• Bruno Santos
• António Pais Antunes (supervisor)

University of Toronto

• Eric Miller

Financial Support

• FCT

Stage of Progress

• Concluded in 2009